Machine Learning Approximations to Predict Epigenetic Age Acceleration in Stroke Patients.

Area of Science:

• Epigenetics and Aging
• Computational Biology
• Vascular Medicine

Background:

• Age acceleration (Age-A) is a predictor of health outcomes, estimated via DNA methylation.
• Age-A is influenced by environmental, lifestyle, and vascular risk factors (VRF).

Purpose of the Study:

• To quantify the contribution of easily measurable factors to Age-A in cerebrovascular disease (CVD) patients.
• To develop an accessible model for predicting Age-A using machine learning (ML).

Main Methods:

• Analyzed a CVD cohort of 952 patients, assessing VRF, lifestyle, and target organ damage.
• Estimated Age-A using Hannum's epigenetic clock.
• Trained six models (linear regression, elastic net, K-Nearest Neighbors, random forest, support vector machine, multilayer perceptron) to predict Age-A.

Main Results:

• Elastic Net (EN) and Multilayer Perceptron (MLP) models demonstrated the best performance.
• Predictive capability was modest, with R-squared values of 0.358 for EN and 0.378 for MLP.
• Identified factors influenced Age-A, but did not explain the majority of its variability.

Conclusions:

• Environmental and lifestyle factors, along with VRF, contribute to Age-A in CVD patients.
• Current easily measurable factors are insufficient to fully explain Age-A variability.
• Further research is needed to identify additional predictors for more accurate Age-A modeling.